Multi-phase, multi-species reactive transport modeling as a tool for system analysis in geological carbon dioxide storage

  • Mehrphasen-Mehrkomponenten-Strömungsmodellierung als Werkzeug für die Systemanalyse geologischer Speicherung von Kohlenstoffdioxid

Naderi Beni, Ali; Clauser, Christoph (Thesis advisor)

1. Aufl.. - Aachen : E.ON Energy Research Center, RWTH Aachen Univ. (2011)
Dissertation / PhD Thesis

In: E.On Energy Research Center / E.ON Energy Research Center <Aachen> 2 : GGE - Applied geophysics and geothermal energy
Page(s)/Article-Nr.: XIV, 143 S. : Ill., graph. Darst., Kt.

Aachen, Techn. Hochsch., Diss., 2011


Geological storage of carbon dioxide (CO2) has been studied worldwide as a possible means for reducing CO2 emissions to the atmosphere. In this context, reservoir modeling which provides both quantitative and qualitative predictions of reservoir behavior is a key element for evaluating a CO2 test injection. Successful implementation of these methods depends on the ability of predicting the physical behavior of the injected CO2 into the subsurface. The better understanding of sequestration and migration processes enables to choose the best site for storage. However, this is not an easy task since coupling of hydrodynamic flow and mass transport in porous media is a very complex physical process. Often phase changes are involved and often the flow is complicated by the presence of chemical species. This thesis contributes to the current efforts to analyze numerically the systems for CO2 underground storage in order to fill some of the scientific gaps identified in this field. Prediction of CO2 plume fate for evaluating CO2 test injections are performed. This is demonstrated in case studies for the Malmö site (Sweden) where a considerable amount of data is available and for the Minden site (Germany) as a potential site with a less than complete existing data set. It is shown that multi-component, multi-phase reactive flow modeling has a high potential for quantifying and identifying different trapping mechanisms and processes in CO2 storage operations. In particular, a few codes allow to account for the different mass transport processes. This turned out to be important for simulations. The variation of the physical properties must not be neglected since it can influence strongly the results. Therefore, a comprehensive data set on seismic, geologic, and geophysical properties form an excellent basis for reservoir modeling. In addition to these data, other parameters must enter the numerical models. Very important are relative permeabilities and capillary pressures, but data is generally sparse. With these caveats and being aware of these constraints, the three-dimensional modeling show that in the Malmö site dissolution trapping mechanism is dominant. However, sensitivity analysis show that the results are in a sensitive range, meaning that a little increase or decrease in a parameter may have a large effect on the results. For example, it is shown how relative permeability and capillary pressure may change the amount and extent of salt precipitation near the injection well ranging from little salt precipitation to a complete well plugging. Further, grid refinement studies imply that simulations on too coarse a grid will overestimate the plume extent. It is indicated that grid block sizes of 0.4 m and smaller are sufficient. Grid resolution appears to be very important. However, it is neglected in many studies and most of the existing models rely on coarse grids. Modeling results show that small inclinations influence the results to some extent, mainly causing some asymmetry of the CO2-phase relative to the injection borehole. It indicates that layer inclinations of less than 2° may be assumed to be horizontal. Non-isothermal, multi-phase flow modeling result also indicates that the temperature variation in question of 5 K at maximum does not alter the fluid and solid material properties significantly and cooling effect due to the Joule-Thomson expansion can be neglected. Reservoir models can help to study the effect of different CO2 injection strategies and to predict the relevant processes. Simulation results indicate that higher injection rates may delay or even inhibit salt plugging. Alternatively, salt precipitation can be reduced or even prevented by practical measures such as injecting fresh water prior to gas. The simulation results of the calcite dissolution experiment indicate that formations that contain mainly carbonate minerals are less suitable for mineral sequestration because calcite dissolves fairly rapidly (on the short-term period of tens of years) which liberates CO2 in dissolution. On the long term, instead, aluminous-silicate reactions dominate resulting in precipitation of a significant amounts of secondary minerals. The reactive transport simulations for the Minden site indicate that after a long time (several hundred years) most of the injected CO2 is fixed in newly formed carbonates such as dawsonite, ankerite, and siderite. It means that hydrodynamic and dissolution trapping mechanisms do not play a role near the injection point and the areas far from the cap rock. Moreover, the estimates of CO2 storage capacity per unit pore volume of the Bunter formation in the Minden site are comparable to those derived from quasi-similar calculations for the Bunter sandstone in the UK sector of the North Sea Basin. The mineral reactions cause a relatively large decrease of porosity and in turn a decrease of permeability in parts of the reservoir. The latter is based on a cubic relationship between porosity and permeability. The modeling results presented here reveal the importance and variations of different processes with respect to many parameters and assumptions to be made in the modeling works and, hence, for further improvements the provided recommendations might be helpful. In this regard and with CO2 storage system analysis, this dissertation serves not only as a feasibility study of the potential sites even with a less than complete data set but also as a guidance for operators. However, they are set up based on existing data. They also are necessarily based on some assumptions such as relative permeability and capillary pressure curves. The performance of these realistic simplified models, e. g. at Malmö, needs to be tested through a series of both field and laboratory experiments. The degree of matching will either confirm current simulations or indicate how modeling should be adjusted. The ultimate aim is to demonstrate that the key modeling assumptions are corroborated by observations. In summary, these numerical simulations of reactive transport provide both a better understanding of the fate of the CO2 plume in a reservoir in time and guidance for practical decisions. This work was supported in part by theWestLB Foundation and E.ON Sverige Värmekraft.